Identification of specific neutralizing antibodies for highly pathogenic avian influenza H5 2.3.4.4b clades to facilitate vaccine design and therapeutics

ABSTRACT The highly pathogenic avian influenza H5 2.3.4.4 and 2.3.2.1c subclades have distinct antigenic properties and are responsible for the majority of human infections. Therefore, it is essential to understand the processes by which antibodies inhibit these subclade viruses to develop effective therapies and vaccines to prevent their escape from neutralizing antibodies. Herein, we report the epitopes of two specific monoclonal antibodies (mAbs) targeting haemagglutinin (HA) of the H5 2.3.4.4b subclade and their neutralizing abilities. The results indicated that the two mAbs provided specific protection against the H5 2.3.4.4b clade viral challenge in MDCK cells and mouse models. Through epitope identification and docking studies, we showed that these novel sites (which are located near the 130-loop (S136, T143) and 190-helix (N199, N205) of HA receptor-binding sites that contribute to the binding affinity of neutralizing mAbs and six residues of the complementarity-determining regions) can be targeted to generate antibodies with enhanced cross-neutralization. This can also help in understanding escape mutations that differ among the H5 2.3.4.4b, h, and 2.3.2.1c subclades. These results provide specific information to facilitate future vaccine design and therapeutics for both subclade viruses, which are dominant and pose a serious threat to humans.

Vaccination is a valuable preventative tool for the current influenza viral infection [6,7].However, owing to the antigenic drift and shift in HA genes, the vaccine is less effective at neutralization and against newer circulating HPAIV H5 clades [8][9][10].
Therefore, the therapeutic use of monoclonal antibodies (mAbs) is a good alternative to vaccines for influenza during the early infectious stages [11,12].
The virus enters the host cell through the receptorbinding sites (RBS) linked to the globular head region of HA via endocytosis, resulting in a pH-dependent entry through endosomal fusion.Viral RNA is then released into the cytoplasm, reaching the nucleus, where it is transcribed for viral replication [13,14].Many mAbs reportedly induce virus neutralization, mainly targeting the RBS of the HA protein of the influenza virus, including the HPAV H5Nx clades [12,15,16].However, most of them targeted the ancestral virus lineages but not the most commonly circulating HPAV H5Nx clades 2.3.4.4b and 2.3.2.1c [15,17,18].Recently, Schuele et al. reported murine monoclonal antibodies that both protect and neutralize the HA of influenza H5 clades 2.3.2.1 and 2.3.4.4; however, the epitope and antibody sequences were not identified in detail [19].
In our previous study, two novel (#11.4 and #23.3) specific mAbs targeting the HA of the H5N6 2.3.4.4b subclade viruses were developed and applied to diagnostic systems [20].Here, we report the neutralization abilities of both mAbs against the H5 2.3.4.4b clade viral challenge in mouse models and Madin-Darby Canine Kidney cell line (MDCKs).Moreover, we found that both mAbs targeted the RBS of the HA of the H5 2.3.4.4b subclade and eight escape mutants on the 130-loop and 190-helix RBS of HA that contribute to the distinct antigenic sites among HPAIV H5 subclades 2.3.4.4b, h, and 2.3.2.1c.Furthermore, we used docking analysis to determine whether parts of the complementarity-determining regions (CDRs) should be engineered to produce antibodies with improved cross-neutralization.  S1), and internal gene segments from A/Puerto Rico/8/34 (H1N1) (PR8) were generated as previously described [20,21].The recommended H5 mutants of A/Anas/KR/2017/2.3.4.4b and RG/A/Swan/MG/ 2020/2.3.4.4 h were generated via site-directed mutagenesis of the plasmids using the primers described in Table S2.Plasmids were transfected into a mixture of MDCK and 293 T cells to rescue viruses.Rescued viruses were amplified in 10-day-old embryonated chicken eggs, and virus stocks were stored at −80 °C.

In vitro microneutralisation assay
In vitro microneutralisation assays were performed as previously described [22].Briefly, 60 µL of serial twofold dilutions of mAbs (160 μg/mL or lower) or positive control (chicken antiserum) were combined with 60 µL of 100 × 50% tissue culture infectious dose per 50 μL (100× TCID50/50 μL) of viruses and incubated for 1 h at 37 °C.The mixture was transferred to a 96-well plate containing 2 × 10 4 MDCK cells per well.After 1 h of adsorption, the viral inoculum was removed, and the plates were washed with phosphate-buffered saline (PBS).The cells were cultivated in Dulbecco's Modified Eagle Medium containing corresponding mAbs and supplemented with 2 µg/mL trypsin.The cells were incubated for 48 h at 37 °C and 5% CO 2 .Subsequently, the supernatant was collected for haemagglutination assay, which was used to determine neutralization.A volume of 50 µL of 0.75% chicken red blood cells was mixed with an equal volume of cell culture supernatant.The resulting mixture was incubated at room temperature for 30 min.Minimal serial dilutions of the positive control that resulted in haemagglutination were used to calculate neutralization titres.For the inhibitory concentration 50% (IC 50 ), agglutination wells were marked, and IC 50 values were calculated using GraphPad Prism Software Version 9.5.0.The experiment was performed in triplicate.

Animal study
The prophylactic and therapeutic efficacies of the mAbs were evaluated using a mouse model following a previously described methodology with modifications [12].Briefly, in the prophylactic study, a group of 5 female BALB/c mice (8 weeks old) received 10, 5, 2.5, or 1 mg/kg of mAbs or 10 mg/kg of control IgG (200 μL) one day before being intranasally challenged with 5 times the 50% mouse lethal dose of A/Anas/KR/ 2017/2.3.4.4b viruses.In the therapeutic investigation, 20, 10, 5, or 1 mg/kg of mAbs or 20 mg/kg of control IgG were intravenously administered into the tail vein 24 h after viral infection.Daily mortality and morbidity rates were recorded.Body weight was assessed for up to 12 d following infection.In the combination-treated group, equal amounts of each antibody were combined to make the same concentration as in the single-treated group.These experiments were performed in conformity with all good laboratory practice recommendations and were authorized by the Wonkwang University Committee on the Ethics of Animal Experiments (approval number: WKU23-52; WKU23-51)

Indirect ELISA
An indirect ELISA was performed as previously described [20,23].Briefly, wild-type and mutated viruses were diluted to 1280 HAU/mL in coating buffer (bicarbonate/carbonate 100 mM, pH 9.6).Further, 100 µL of each virus was coated on a 96-well microtitre plate (Greiner GmbH, Pleidelsheim, Germany) and incubated overnight at 4 °C.The plates were rinsed with PBS containing 0.1% v/v Tween-20 (PBST) and blocked in PBST plus 5% non-fat milk for 2 h at 37 °C.For virus detection, 20 µg/ mL of primary #11.4 or #23.3 mAbs and 0.2 μg/mL 100 μL/well of secondary IgG-HRP (Catalog Number: ab97046; Abcam, Cambridge, UK) was added and incubated for 1 h.After washing five times to remove nonspecific binding, 100 μL of WANTAI BioPharm tetramethylbenzidine substrate was added at room temperature for colour development for 10 min; subsequently, the reactions were stopped by adding 100 μL of 0.18 M sulphuric acid to each well.Optical density was measured at 450 nm using a microplate reader (SpectraMax® M Series Multi-Mode Microplate, San Jose, CA, USA).

Monoclonal antibody sequencing and docking study
Monoclonal antibody variable sequences were obtained using SMART technology (https://www.takarabio.com/learning-centers/mrna-and-cdna-synthesis/cdna-synthesis/cloning-antibody-variable-regions) as previously described [24].Mouse hybridoma clones were cultured and confirmed using an IsoStrip™ Mouse Monoclonal Antibody Isotyping Kit (11493027001; Roche Diagnostics, Rotkreuz, Switzerland).Total RNA was extracted from 10 6 cells using a Macherey-Nagel™ NucleoProtect RNA kit (Macherey-Nagel™ 740400.50;Fisher Scientific, England, UK).Specific PCR primers and thermal conditions were selected based on a previous study.PCR products were cloned into a pCR®-Blunt II-TOPO® vector using the Zero Blunt™ TOPO™ PCR Cloning Kit and analysed via Sanger sequencing.The DNA sequence of each antibody-variable region was analysed using IgBLAST with default parameters, and the query organism was set to mouse (https://www.ncbi.nlm.nih.gov/igblast/) and SAbPred (https://opig.stats.ox.ac.uk/webapps/sabdab-sabpred/sabpred/abod ybuilder2/).This analysis was conducted to determine the amino acid sequences and percentage identity between the antibody-variable regions and reference sequences provided by IgBLAST for both light and heavy chains.To get the 3D-structures of antibodyvariable regions, the final amino acid sequence for each antibody was uploaded to the I-TASSER server (https://zhanggroup.org/I-TASSER/) (Table S3).Further, protein-protein docking was performed using the LZerD Web Server with epitope site constraint and adjusted parameters for residue-residue distance (0 to 3 Å) (https://lzerd.kiharalab.org/upload/).The PyMOL Molecular Graphics System was used to view residues that interacted with mAbs and antigens (https://pymol.org/2/).

In vivo prophylactic and therapeutic efficacy of mAbs in mouse model
We investigated the prophylactic and therapeutic efficacies of both mAbs against RG/A/Anas/KR/ 2017/2.3.4.4b (H5) viral infections in a mouse model.Groups of 8-week-old female BALB/c mice (n = 5) were intranasally administered five lethal doses (LD 50 ) of RG/A/Anas/KR/2017/2.3.4.4b (H5) viral infection in a mouse model, and antibodies (dosage range of 1-20 mg/kg) were administered 24 h before and after viral infection for the prophylactic and therapeutic groups, respectively.The results showed that in the prophylactic groups (n = 5), doses of at least 2.5 mg/kg of both mAbs could completely protect mice from a lethal infection with RG/A/ Anas/KR/2017/2.3.4.4b (H5).Further, these doses of #23.3 mAb had less impact on mouse body weight compared with the #11.4 mAb (Figure 2a, c).In addition, doses as low as 1 mg/kg provided 60% survival for both #23.3 and #11.4 mAbs.In the therapeutic groups (n = 5), 5 mg/kg and above of #11.4 mAb could fully protect the mice against the RG/A/ Anas/KR/2017/2.3.4.4b (H5) virus, whereas a higher dose of 10 mg/kg and above of #23.3 mAb provided similar protection against a lethal infection with RG/A/Anas/KR/2017/2.3.4.4b (H5) virus.The lower dosage of 2.5 mg/kg of both #23.3 and #11.4 mAbs offered partial protection, with a survival rate of 40% and 60%, respectively (Figure 2b, d).The administration of #23.3 mAb to therapeutic groups exhibited considerable changes in body weight, which were less pronounced than those in the prophylactic groups.Therefore, in vivo prophylactic and therapeutic results indicated that both #23.4 and #11.4 mAbs could protect mice against the RG/A/Anas/ KR/2017/2.3.4.4b (H5) virus, although this protection was less effective in the therapeutic groups than in the prophylactic groups.We next investigate whether the combination treatment with #23.3 and #11.4 has any coaction to protect mice against the RG/A/Anas/KR/ 2017/2.3.4.4b (H5) virus (Figure 2e, f).In the prophylactic groups, treatment of mice with combined #23.3 and #11.4 mAbs (lower dosage of 1 mg/kg) resulted in an 80% survival rate.It increased the survival rate by 20% in comparison to a single treatment at the same dosage (1 mg/kg) of #23.3 and #11.4 (60%).In the therapeutic groups, this survival rate was obtained at 60% and 100% at lower dosages of 2.5 and 5 mg/kg, respectively.It was similar to that obtained with #11.4 monotherapy at dosages of 2.5 mg/kg (60%) but was higher than that obtained with #23.3 monotherapy at lower dosages of 2.5 mg/kg (40%) and 5 mg/kg (80%).In conclusion, combination therapy with both mAbs has some advantages in protecting mice on prophylaxis.However, in therapeutic treatment, combination therapy did not show any effectiveness compared to single treatment with mAbs for improving mouse survival rates.

Epitope selection for validation via reverse genetics of influenza virus
To determine the epitopes of #23.3 and #11.4 mAbs, we compared the amino acid sequences among three HPAIV H5 subclades, including RG/A/Anas/KR/ 2017/2.3.4.4b,RG/A/Swan/MG/2020/2.3.4.4 h, and RG/A/VN/2014/2.3.2.1c (Figure 3).Because both mAbs showed specific neutralizing activities against RG/A/Anas/KR/2017/2.3.4.4b, we selected three potential epitope regions (B5, H3, and B6) that belong to the RBS of HA, with high genetic variants among the three subclades.For comparison, we also selected a B8 region that did not belong to the RBS to generate mutated viruses via reverse genetics (RG).The site mutation of the epitopes was introduced via mutagenesis of plasmids following changes in amino acid sequences and primers (Table 1 and Table S2).The plasmids were then transfected into mammalian cells to rescue the RG-mutated viruses.RG viruses induced a cytopathic effect 4 days post-transfection (Figure S3).Finally, four mutated viruses were generated, and their sequences were confirmed using Sanger sequencing (http://www.cosmogenetech.com/main.jsp).

Epitope validation via indirect ELISA
To validate the binding affinity of the antibody candidates to mutated viruses, we performed an indirect ELISA to compare wild-type (Wt) and mutated (Mut) viruses (Figure 4).The results showed that #23.virus clades were different, and the epitopes of #23.3 and #11.4 mAbs were far from each other (Figure 5a).In addition, alignment of epitope amino acid sequences showed that #23.3 mAb completely belonged to the 190-helix, whereas #11.4 mAb belonged to the 130-loop, partially in the RBS region (Figure 5b).Altogether, these results indicate that the amino acid positions N199S, T204I, and N205S seem to be key elements that contribute to the binding affinity of #23.3 mAb, and that S136D, P139S, N140D, and T143A contribute to the interaction between #11.4 mAb and the HA antigen target.

Antibody sequencing and docking study
DNA sequences corresponding to the variable region of the immunoglobulin light chain (VL) and heavy chain (VH) of samples #23.3 and #11.4 mAbs were obtained and compared with the nearest germline sequence using the IgBLAST database.These findings indicated that the sequence labelled #23.3 mAb corresponded to the specific germline IGKV4-74*01 and IGHV1S81*02 V genes for VL and VH, respectively.The nucleotide identity of the VL chain was 99.3% (290/292 nucleotides), whereas that of the VH chain was 98% (288/294 nucleotides).In addition, the alignment of #11.4 mAb corresponded to a specific IGKV9-120*02 for VL and IGHV1-63*02 germline V gene for VH, with a nucleotide identity of 99.6% (283/284 nucleotides) and 94.9% (279/294 nucleotides), respectively.We then performed protein-protein docking to predict the interaction between the mAbs and antigen target (H5N6/HA1/A/Anas/KR/2017/ 2.3.4.4b).The results showed that #23.3 mAb interacted widely with the antigen target, creating eight hydrogen bonds located in the RBS of the antigen, including seven contacts of HC-CRD1:CDR3 of the mAb with the 190-helix (N199, A200, E201, and N205) of the antigen target (Figure 6a and Table 2).In addition, only one hydrogen bond of LC/CDR2-S241 was in contact with the antigen residue D171.For #11.4 mAb, HC/ CDR3, LC/CDR1, and LC/CDR3 formed four hydrogen bond contacts with two residues, S136 and T143, on the RBS of the antigen target (Figure 6b and Table 2).Interestingly, the S136 residue near the 130-loop was the main interaction, accounting for 75% (3/4) of the total hydrogen bonds.Generally, the hydrogen bonds of both mAbs with antigen targets were strong within 2.7 angstroms (Å) between the donor-acceptor atoms and binding sites on the RBS of the antigen target [25].

Discussion
The spread of AIVs among wild birds has potential consequences for human health and the global poultry industry [26][27][28][29][30]. HPAIV H5 2.3.4.4 and 2.3.2.1c subclade infections in animals have jumped the species barrier and spread to humans [31][32][33].To effectively battle HPAIV H5, universal vaccine candidates or broadly neutralizing antibodies targeting conserved epitopes are necessary [6,12,34].However, the antigenic evolution of influenza viruses and HPAIV H5 subclades is a substantial barrier to the research and development of antiviral medicines and universal vaccines with high efficacies and long shelf lives [35].Consequently, a therapeutic window using specific neutralizing antibodies would be useful for the current emergence and circulation of HPAIV H5 strains.
In this study, we analysed the biofunctionality and structure of #11.4 and #23.3 mAbs that specifically neutralized and targeted HA in H5 2.3.4.4b subclade viruses.We also analysed the key amino acids that contributed to the binding between mAbs and protein targets.In particular, both mAbs displayed significant neutralizing efficacy against strains of HPAI H5N1 2.3.4.4b subclade viruses in both in vitro and in vivo testing, and the binding sites of the mAbs may be located on the 130-loop and 190-helix RBS of the HA protein of the H5 2.3.4.4b subclades.
In a previous study on a cross-neutralizing antibody that targeted the HA of H1N1 and H5N1, the results revealed that IC 50 values of 5.75 μg/mL and a dosage of 10 mg/kg or above of mAbs could completely protect mice from lethal infection of viruses in both prophylactic and therapeutic groups [12].Schuele et al. found that mAbs completely neutralized the HPAI H5 2.3.4.4 and 2.3.2.1 subclades at concentrations between 0.3 and 20 μg/mL [19].In our study, we obtained IC 50 values of 0.33 and 0.36 µg/mL for #23.3 and #11.4 mAbs, respectively, and also found that doses of 10 mg/kg or higher of mAbs protected the mice from deadly viral infection.Therefore, our antibody is superior or comparable to previous reports   S3.
on protection against and neutralization of viruses in mice.
The identification of mAb epitopes can be accomplished using various approaches, including crystal structure analysis, escape mutation, random mutagenesis library, and site-direct mutagenesis [36].In this study, mAb epitopes were validated using site-directed mutagenesis and RG of influenza virus via indirect ELISA.The RBS structure is characterized by a cluster of conserved amino acids which are bordered by the 190-helix, 130-loop, and 220-loop [13].Therefore, both mAbs were expected to directly block viral binding.However, additional experiments should be conducted to determine the binding mechanism of mAbs to the HA of H5 viruses.
Multiple studies have demonstrated that mAb epitopes that specifically target RBS have the potential to directly inhibit viral binding [11,12,16,18,19].Five positions (S124, P127, N128, E130, and T131) that contribute to the binding affinity of #11.4 mAb overlap with the 130-loop RBS of the HA of the H5 2.3.4.4b subclade, such as C12H5, FLD21.140,65C6, 3C11, and AVFlulgG01 mAbs, which reportedly neutralize ancestor virus lineages that include HPAIV H5; 2, 8, 7.1, and 1 clades [12,15,36].Unlike #11.4 mAb, the binding affinity of #23.3 mAb is influenced by three specific positions (N190, T195, and N196) within the 190-helix.These positions exhibit significant spatial separation from the binding sites of #11.4 mAb, thereby precluding any epitope competition between the two mAbs.Furthermore, mutations occurring at these sites within the 190-helix did not alter the binding affinity of #11.4 mAb.Thus, the combined administration of our mAbs showed better efficacy than single treatment of mAbs against viral infections in prophylactic studies.This result implies that two mAbs recognize spatially distinct HA antigen domains, demonstrating modest complementarity in terms of their functionality to block viral entry.Therefore, the mAbs that recognize different antigen targets should be further investigated in order to determine the most effective and robust approach for preventing viral infection.Docking studies involve the use of computational tools to predict binding sites within the complex structure of an antibody and antigen target or other molecules, thereby facilitating the identification of potential interactions [37][38][39][40].In this study, the docking results suggested 12 potential interactions between CDRs and seven binding residues on the antigen target.Among these, HC/ CDR1-S50 and HC/CDR3-D121 of #23.3 mAb, and HC/CDR3-E121, LC/CDR1-S213, CDR3-Y277, and LC/CDR3-S275 of #11.4 mAb may contribute to the binding specificity of mAbs because of their contact with non-conserved amino acids in 2.3.4.4b, h, and 2.3.2.1c, including N199, N205, S136, and T143.Therefore, the abovementioned sites on CDRs may serve as potential therapeutic targets to facilitate the development of modified antibodies with cross-neutralizing capabilities.
In addition to the significance of this research, there are some limitations that should be considered in this study.At first, this study only looked at a small number of specific neutralizing antibodies and didn't test to see if these sites were enough to make antibodies that could neutralize HPAIV H5.Therefore, the immunogenicity of escape mutations or incorporation of other parts should be validated before designing vaccines against H5Nx 2.3.4.4 and 2.3.2.1c subclades.The second restriction relates to the six amino acid residues of the CDRs, as anticipated by docking studies.Does this have a positive impact on the generation or modification of antibodies?To answer this question, one recombinant antibody (#23.3) was successfully produced and verified utilizing the E. coli expression method (data not shown in this study).After that, changes will be made to the CDRs to improve the antibody's affinity or the mAbs' ability to cross-neutralize.
In conclusion, we identified a major mutation site in the RBS of H5 that contributes to viral escape from neutralizing antibodies among the different subclades 2.3.4.4b, h, and 2.3.2.1c.We found eight key points (S136, P139, N140, E142, T143, N199, T204, and N205) which may be useful during virus surveillance activities, for designing improved vaccines for the HPAIV H5Nx clades, particularly the 2.3.4.4b subclade, and for engineering antibodies that can efficiently destroy viruses and identify escape variants.

Figure 4 .
Figure 4. Epitope identification of specific neutralizing antibodies for highly pathogenic avian influenza H5 2.3.4.4b clades of the #23.3 and #11.4 mAbs via indirect ELISA.The means and standard deviation (SD) were calculated, and Student's t-test was performed using GraphPad Prism Software Version 9.5.0 (La Jolla, CA, USA).Results are presented as the means ± SD. *** p < 0.001; ns, not significant; P.C., positive control anti-influenza A nucleoprotein.

Figure 5 .
Figure 5. 3D-modelling and alignment of epitopes of specific neutralizing antibodies on the HA head domain of three H5 virus clades.(a) Key residues of #11.4 and #23.3 mAbs on the HA of H5 viruses; the blue and red regions show all epitopes of #11.4 and #23.3 mAbs, respectively; key amino acid types are indicated with bold colours.(b) Alignment epitopes among three different H5 virus clades; numbering of the receptor binding sites with the 130-loop, 190-helix, and 220-loop are based on the HA of H3.

Figure 6 .
Figure 6.Prediction of hydrogen bonding interactions between antibody and antigen of the H5N6/HA1/A/Anas/KR/2017/2.3.4.4b clades using docking LZerD Web Server (https://lzerd.kiharalab.org/upload/).(a) Hydrogen bonding interactions of #23.3 mAb and HA1/2.3.4.4b clade; the heavy chain (HC) and light chain (LC) of #23.3 mAb are shown in bright orange and green colours, respectively.(b) Hydrogen bonding interactions of #11.4 mAb and HA1/2.3.4.4b clade; HC and LC of #11.4 mAb are shown in bright pink and blue colours, respectively.The probable hydrogen bonds are shown as black dashed lines.The binding sites of CDRs are highlighted in deep colours at both HC and LC of the mAb, and antigens sites are shown in bright grey colour.The mAb sequences for 3D-modelling are shown in TableS3.

Table 1 .
List of reverse genetics viruses for validation of the epitopes of mAbs.